using Machine Learning and Density Functional Theory*

Lead-free metal halides have potential uses in photovoltaic and other optoelectronic applications. AgSbI₄ is an example of a lead-free metal halide that has recently been synthesized but not yet fully treated with theoretical calculations due to the large degree of site disorder in its cation sublattice. We employ a machine learning model of the total energy in AgSbI₄ to choose a few simulation cells out of ~10⁷ possibilities; a task out of reach for current first principles methods. We calculate the structural and optoelectronic properties such as X-ray diffraction patterns and band gaps of these cells and find that they agree well with existing experimental data. Additionally, new predictions of the optoelectronic properties indicate that AgSbI₄ may be a good choice for use as an absorber layer in tandem photovoltaic cells.